It is often necessary to visually evaluate a screen of a mobile device (e.g., a smartphone or tablet) to identify cracks or other defects in the mobile device. For example, pricing the mobile device, assessing the mobile device for possible repair, and evaluating the mobile device for warranty coverage all may require identification of any cracks in the mobile device's screen and/or in non-screen portions of the device. Individualized manual inspection of mobile device screens for cracks is slow, cumbersome, and can yield inconsistent results. Current automated methods for detecting cracks in other contexts are often over-inclusive resulting in high rates of false-positive crack indications. Accordingly, there is a need for improved methods and systems for automatically detecting cracks in mobile device screens.